A single view of risk fuels further success for Murex

Asia-Pacific (Apac) banks are feeling myriad pressures that demand change and evolution.

Clients expect a steady flow of new structured products. Regulatory timetables and demands continue to diverge. Bouts of volatility have pushed up structured product volumes, and the risk calculations behind them are straining hardware that was deemed adequate a few years ago. Meanwhile, artificial intelligence tools are helping attackers find software vulnerabilities faster than many institutions can patch them.

Murex pairs its cross-asset platform, MX.3, with services and solutions that help its large and diverse Apac client base – banks, corporate treasuries, neobanks and commodities firms – meet these demanding challenges.

MX.3 provides one view of risk 

Trading desks in the region increasingly hold positions that generate risk across several asset classes at once, such as foreign exchange, rates, equity and commodity risk. Consistently pricing and hedging those correlated exposures is difficult when each asset class sits on a separate system with its own models. Some 85% of Murex’s Apac clients run three or more asset classes on MX.3. A similar percentage added new modules or products to their installations in the year to September.

LiveBook – MX.3’s real-time position monitoring, risk calculation and profit‑and-loss (P&L) engine – is designed with that problem in mind. It delivers live P&L, Greeks and position updates across asset classes, and replaces end-of-day reports with a continuously refreshed view of risk.

Keeping large exotic books ticking in real time also raises a question of cost. This year, Murex extended its risk-based P&L framework, which uses a Taylor expansion to estimate P&L and sensitivities between full revaluations. New risk-based DV01 and vega measures capture convexity effects, while dashboards track the gap between the estimate and periodic full revaluation. For rates books with hundreds of risk factors, MX.3’s principal component analysis-based gamma framework reduces exposure to three drivers – level, slope and convexity – which explain more than 95% of variance.

“When correlations move across rates, FX, credit and commodities at the same time, desks can’t afford to reconcile separate systems or approximate their hedges,” says Matthieu Avanthey, head of functional product for Apac at Murex. “Our clients priced, managed and rebalanced in one place, in real time – one framework, one pricing engine and one risk view.”

Trading desks equipped with MX.3 enjoy a competitive advantage: a system that natively enables launching new products or even new business lines.

Sharp price swings in Australia’s power market have pushed corporates to hedge electricity exposure, and one Australian bank launched an electricity desk on MX.3 in late 2025, with pricing and settlement built around the national market’s five-minute settlement rule. It is now adding exotic payoffs.

In Japan, a trading house using Murex’s software-as-a-service offering was able to trade newly listed Japanese power futures within weeks of its launch.

Murex has also delivered sharia-compliant structured products for Islamic banking in the region.

Structured products under strain

Demand for structured products across Asia has held up through the volatility. At the same time, the products are getting more complex and harder to run at volume.

In rates, Malaysia’s move from the Kuala Lumpur interbank offered rate to the Malaysia overnight rate is landing on high-volume callable range-accrual books, with no standard fallback to lean on. Banks in India and Thailand have launched callable notes linked to compounded overnight rates and callable reverse floaters. A Taiwanese bank used MX.3’s Python extension framework to bring constant maturity swaps range accruals to market within months. When volumes outgrew that setup, Murex converted the most frequently traded structures into native payoffs, lifting performance without extra hardware.

On the equities side, five Apac derivatives desks have selected MX.3 over the past 12 months. The firm’s Light Exotics module lets users build payoffs without code and picks the pricing model automatically. One Singapore bank is rolling out more than 40 variants on it.

Meanwhile in FX, USD/JPY volatility has revived demand for target redemption forwards among Japanese corporates. One Japanese regional bank put seven variants into production across pricing, credit risk, margining and regulatory reporting, with Murex’s validation team helping it through model approval. The products run on a partial differential equation-based local volatility model that prices four times faster than Monte Carlo.

Calculation loads are rising with every new regulatory scenario and intraday risk run, and conventional pricing methods are reaching their limits. With machine learning, Murex believes complex derivatives pricing models can be replicated with high accuracy at a speed that makes them practical for real-time, high-volume use.

For stochastic local volatility (SLV)calibration, Murex’s research and development (R&D) team has already developed a neural network calibration layer for TStar, its flagship SLV model. Trained on 220 million prices, it enables calibration at vanilla pricing speed without graphics processing units, meaning risk matrices that took hours now finish in minutes.

For Murex, change is routine

Newly disclosed software vulnerabilities across the industry were up about 75% in the second quarter of 2026 compared with a year earlier, according to CVE.ICU data, and AI tools have closed the gap between discovery and exploitation by bad actors. In response, Murex has nearly doubled its security teams over two years and builds automated controls and penetration testing into each release; a critical HTTP/2 vulnerability found in June was patched in the main MX.3 release within four days.

Getting these fixes into a bank’s production is much slower because every upgrade must be tested against the bank’s own customised, tightly interconnected setup. Known flaws can stay in production for months or years. Banks are also keen to harness the potential of rapidly evolving technology, such as the tokenisation of capital markets flows or agentic AI applications.

Murex consistently reinvests 18% of its turnover in R&D to keep its MX.3 platform current. The challenge, however, is how quickly financial institutions can bring these updates into their own IT environments. The slow speed of evolution across their IT landscapes has become a strategic obstacle.

MXevolve, Murex’s upgrade-as-a-service offering, largely solves this equation. Murex builds automated tests around each client’s configuration and data, executing them against every quarterly release. Problems surface early, significantly reducing upgrade effort and operational risk. Clients on the service run versions with a median age of 18 months – against more than four years for other clients – reducing exposure to security vulnerabilities, regulatory gaps and defects already addressed in newer releases.

“Banks have tended to treat an upgrade as a major project every few years. That doesn’t work when a security fix is needed now,” says Alexandre Bon, Apac head of market development at Murex. “What we’re trying to do is make change routine.”

Tokenisation is progressing towards institutional use cases. In parallel, generative and agentic AI are moving from productivity and user assistance towards supporting workflows and controlled actions across the trade lifecycle. Financial institutions must now embrace these innovations without creating new silos, while preserving trusted data, risk management, security, human oversight and governance.

MX.3 evolves with these needs: integrating digital assets into existing trading, risk, collateral and post-trade processes while extending connectivity to new market infrastructures. Murex is also investing in governed agentic AI grounded in MX.3’s business context, positions, calculations, processes and controls. Through an open model-agnostic agentic framework, users and agents interact with trusted data and services while keeping outcomes reliable, explainable and controlled. Its first client-facing AI agents are due over the next year, starting with natural language trade booking, confirmation matching and processing, and P&L explain and analysis.

“We are delighted our hard work and innovation continues to be recognised in the Risk Asia Awards,” says Nick Thomas, Apac regional head at Murex. “What sets us apart is something our clients tell us directly: beyond a platform that works today, they want a partner that keeps them current tomorrow. Vulnerability discovery is accelerating, geopolitics and regulations are compressing timelines, and product complexity keeps rising. Our job is to make sure none of that slows them down.”

スポンサーコンテンツ